The present invention relates to a computer system using a disk controller of a disk array for storing data into a plurality of disk devices and an operating service thereof.
Currently, a data center or the like handling a large volume of data introduces various storage devices such as a large disk array connected to a mainframe as well as a small and medium disk array connected to a tape backup device or an open system servers, and stores information in each of the devices. At present, it is hard to say that the stored information is organically strongly linked to the respective devices. Thus, active has been a movement to introduce a concept SAN (Storage Area Network), and to connect various storage devices through the network, thereby managing information. A large data center has used a large disk array to build a storage system which has high reliability, high performance and high functionality. With SAN, a larger storage system than ever is being built in addition to a small and medium disk array.
On the other hand, in the future, a small and medium disk array will be required to have high performance and high reliability. A current disk array will need scalably supporting from a small storage system such as SOHO to a large storage system such as a bank, and also a service capable of operating this effectively.
A prior art disk array centers on a disk controller as shown in FIG. 1. The disk controller has a plurality of channel interface (hereinafter, channel IF) parts 103 for executing data transfer between a host computer 101 and a disk controller 109, a plurality of disk interface (hereinafter, disk IF) parts 104 for executing data transfer between a magnetic disk device 102 and the disk controller 109, a cash memory part 107 for storing data read and written between the channel IF and the disk channel IF, and a common memory part 108 for storing control information on the disk array controller 109 (for example, control information on data transfer between the channel IF part 103 and the cash memory part 107). The channel IF part 103, the disk IF part 104 and the cash memory part 107 are connected by a mutual connecting network 105. The channel IF part 103, the disk IF part 104, and the common memory part 108 are also connected by a mutual connecting network 106. The mutual connecting network herein means all connection means such as a switch, loop, and bus. Herein, the numeral 412 denotes an SVP (Supervise Processor) part and, as described later, collects information on the access number per each channel path number of information transmission between the host computer 101 and the disk controller 109.
When one disk controller is formed by such a construction, scalability is a form of adding a component to its minimum construction as a basic device, i.e., of sequentially adding an optional component for extension. In other words, its minimum construction needs to have an extension mechanism for adding an optional component up to its maximum construction, and its small construction is also provided with a mechanism required for extension. They are unnecessary mechanisms when operated by its basic construction. The device cost can be inevitably relatively high for its basic construction. Further, to meet speeding up of a host computer and improvement in connectivity (an increase in the number of connectable host interfaces), it is also necessary to respond to speeding up of the connection mechanism for the extension components and improvement in extensionality (an increase in the number of the extendable components). The cost is increased, so that there is a high possibility that the cost may be relatively high for its basic construction.
With respect to this, as shown in the overview of FIG. 2, an SAN (Storage Area Network) environment using a disk array makes a system construction efficient. Host computers 101 are connected through a common mutual connecting network 210 to a disk controller 109. The disk controller has a plurality of channel IF parts 103, a plurality of disk IF parts 104, a plurality of common memory parts 107, and a plurality of cash memory parts 108, which are connected to mutual connecting networks 105 and 106, respectively. The disk IF part is connected to a plurality of disk devices 102. The disk controller 109 and the disk device 102 connected thereto function as a disk array. The common mutual connecting network 210 can connect many kinds of storage devices, and can also connect a magnetic tape memory 212. Specifically, there are considered every kind of network such as a fiber channel switch, loop and LAN. In such a form, for example, tens of or hundreds of small disk array devices are connected to from an aggregate of a large number of logical volumes, which can show the system to an upper class host computer. It is possible to realize a large volume equal to that of a prior art large disk array with high availability and high reliability as well as high function such as logical volume copy between the disk controllers, as shown by a path 211 in the drawing. However, there is the problem that the small disk array device does not pursue high availability and high reliability of the large disk array. On the other hand, there is a merit in cost because of the aggregate of inexpensive small disk arrays. Also herein, the numeral 412 denotes an SVP (Supervise Processor) part, which collects information on the access number per each channel path number of information transmission between the host computer 101 and the disk controller 109.
As shown in FIG. 3, as a similar construction, a unit disk controller 309 functions as a disk array by consisting of a disk controller provided with mutual connecting networks 105 and 106 to connect a channel IF part 103 and a disk IF part 104 for connecting host computers 101 and disk devices 102, respectively, a common memory part 107, and a cash memory part 108, and is smaller than the disk controller shown in FIGS. 1 and 2. A plurality of the unit disk controllers can be connected by a common mutual connecting network 310 so as to construct a disk array functioning as a disk controller 315 on the whole. In this case, the unit disk controller is about ¼ to ½ times the prior art disk controller and is mounted in a compact manner. The cost can be thus low. Further, the common mutual connecting network 310 as its center part has a minimum band required. The cost of the entire device can be reduced. Also herein, the numeral 412 denotes an SVP (Supervise Processor) part, which collects information on the access number per each channel path number of information transmission between the host computer 101 and the disk controller 315, more strictly, all the unit disk controllers 309.
There can be a system integrating the system construction shown in FIGS. 2 and 3, although the illustration thereof is omitted. In other words, the host computers 101 in FIG. 3 are connected by the common mutual connecting network 210 shown in FIG. 2. The host computer not through the unit disk controller directly connected to itself can access a unit disk controller directly connected to other host computer. Consequently, it is unnecessary to pass through the mutual connecting network between the unit disk controllers, whereby access can be improved.